At present, a high dynamic range image may be acquired by exposing a plurality of pixels of an image sensor synchronously with three different exposure times. However, in this way, the three different exposure times are calculated according to a histogram of the whole image Since the whole image does not has a linear brightness, calculating the three exposure times with the histogram of the whole image cannot realize a better AE (auto exposure) convergence.